Hi,
on monday I updated my F12 the last time including the updates-testing repo. Since then my computer switches of before reaching the gdm login. I tried to boot without starting X and I also tried other kernels. Still all the same, computer switches off.
When I read the test-list I saw that Bruno Wolff was having a problem that sounded like mine. Probably udev is broken.
But what can I do now to fix this? I do not want to reinstall fedora... is it possible to do something with a liveCD that would fix udev?
Stefan
On Thu, 2010-03-25 at 08:04 +0100, Stefan Grosse wrote:
Hi,
on monday I updated my F12 the last time including the updates-testing repo. Since then my computer switches of before reaching the gdm login. I tried to boot without starting X and I also tried other kernels. Still all the same, computer switches off.
When I read the test-list I saw that Bruno Wolff was having a problem that sounded like mine. Probably udev is broken.
But what can I do now to fix this? I do not want to reinstall fedora... is it possible to do something with a liveCD that would fix udev?
Log in to liveCD, mount your f12 partition somewhere, mount --bind the current /proc to it, chroot to it and do yum update *udev*. The new udev fixed it. (I've done this exact procedure, albeit from F13 instead of LiveCD.) The udev version that works for me is:
udev-145-19.fc12.i686
Martin
On Thu, Mar 25, 2010 at 09:08:32 +0100, Martin Sourada martin.sourada@gmail.com wrote:
Log in to liveCD, mount your f12 partition somewhere, mount --bind the current /proc to it, chroot to it and do yum update *udev*. The new udev fixed it. (I've done this exact procedure, albeit from F13 instead of LiveCD.) The udev version that works for me is:
udev-145-19.fc12.i686
That's pretty much what I did as well. I also bind mount /sys and /dev and mount other filesystems (such as /boot). For luks encrypted partitions you need to use cryptsetup luksOpen before you can mount. You might also need to mess with your network settings if you use static IPs.
Other people had success by adding a kernel parameter to run a shell instead of the normal init and then either blacklisted the watchdog timer or modified a udev rule to prevent the reboots.